Improving Entity Linking using Surface Form Refinement

摘要: In this paper, we present an algorithm for improving named entity resolution and linking by using surface form generation rewriting. Surface forms consist of a word or group words that matches lexical units like Paris New York City. Used as matching sequences to select candidate entries in knowledge base, they contribute the disambiguation those candidates through similarity measures. context, misspelled textual (entities) can be impossible identify due lack available forms. To address problem, propose refinement based on Wikipedia resources. The approach extends coverage our system, rewrites reformulates mentions prior starting annotation process. is evaluated corpus associated with monolingual English task NIST KBP 2013. We show improves system performance.